Contents lists available at ScienceDirect

Journal of Membrane Science

journal homepage: www.elsevier.com/locate/memsci

Highly permeable nanoporous block copolymer membranes by machine-casting on nonwoven supports: An upscalable route

Xiansong Shi, Zhaogen Wang, Yong Wang⁎

State Key Laboratory of Materials-Oriented Chemical Engineering, College of Chemical Engineering, and Jiangsu National Synergetic Innovation Center for AdvancedMaterials, Nanjing Tech University, Nanjing 210009, Jiangsu, PR China

A R T I C L E I N F O

Keywords:Block copolymerMachine castingSelective swellingComposite membraneFractionation

A B S T R A C T

Block copolymer (BCP) membranes are distinguished for their well-defined porosities, tunable pore geometries,and functionable pore walls. However, it remains challenging to produce robust BCP membranes by affordable,convenient methods. Herein, we demonstrate a facile and easily upscalable approach to produce highlypermeable BCP membranes in large areas. The membranes possess a bi-layered composite structure withnanoporous polystyrene-block-poly(2-vinylpyrdine) BCP layers directly supported on macroporous nonwovensubstrates. The BCP layers are machine-cast on the water-prefilled nonwoven, and interconnected nanoporositiesare created in the BCP layers by ethanol swelling. The nanoporous BCP layers exhibit a thickness of ~10 µm andare tightly adhered to the nonwoven. Changes in the swelling temperatures and durations modulate both poresizes and surface hydrophilicity of the BCP layers, and consequently the permselectivity of the membranes. Byincreasing swelling duration from 15 min to 12 h, the permeability of the membrane swollen at 65 °C can beincreased from ~100 to ~850 L m−2 h−1 bar−1 with the retention to 15-nm gold nanoparticles reduced from~93% to ~54%. Moreover, we demonstrate that the composite membrane can efficiently fractionatenanoparticles and narrow down their size distribution from ~3–20 nm to ~3–10 nm.

1. Introduction

Membrane separation is playing an increasingly significant role inindustry as well as daily life for their wide applications in dynamicfields such as water treatment [1], biotechnology processes [2], andhealth care [3]. To improve separation performances and also toexpand the applications of membrane separation in new areas, it isparticularly attractive to fabricate new membranes or modify existingmembranes. New materials, for example, carbon nanotubes [4,5],graphene and its derivatives [6,7], metal-organic frameworks [8,9],and self-assembled polymers [10,11] have been used to prepareadvanced membranes with fast permeability and/or sharp selectivity.Among them, block copolymers (BCPs) are of particular interestbecause they can readily microphase-separate into highly orderednanoscopic structures leading to homogeneous membrane pores[12–14]. Moreover, BCPs-based membranes enjoy the merit of avail-ability of membrane pores with different geometries and the flexibilityin functionalization of membrane pores. Selective etching/extraction ofthe minority phases [15–17], nonsolvent-induced phase separation(NIPS) of concentrated BCP solutions [18–20], and selective swellingof amphiphilic BCPs [21–23] are the three main strategies for the

preparation of nanoporous membranes from BCP precursors. However,BCP membranes are currently suffering from several challenges includ-ing high cost of the BCP precursors, the weak mechanical stability, andcomplicated manufacturing process, which severely limit their upscal-ability and real-world applications. To tackle the issue of high cost andalso mechanical strength, the bi-layered composited membrane struc-ture consisting of a thin BCP selective layer on the top of a macroporoussubstrate is typically used [24]. Such an approach allows to design andprepare the two layers independently with the nanoporous thin skinsensuring high selectivity at small consumption of BCP raw materialsand the underlying macroporous substrates promising strong mechan-ical robustness and low flow resistance. Recently, a number of methodshave emerged for the preparation of composite membranes having ananoporous BCP selective layers. Transferring spin-coated BCP thinfilms onto macroporous substrates is a versatile method to produceultrathin layers of various BCPs [25–27]. However, it is very tediousand is difficult to produce large-area membranes. Alternatively, directcoating dilute BCP solutions onto the surface of liquid-filled poroussubstrates is a much convenient process capable of making BCP layerswith the thickness down to a few micrometers directly on the substrate[16,28,29]. Hillmyer et al. prepared composite membranes by manu-

http://dx.doi.org/10.1016/j.memsci.2017.03.046Received 14 January 2017; Received in revised form 27 March 2017; Accepted 28 March 2017

⁎ Corresponding author.E-mail address: yongwang@njtech.edu.cn (Y. Wang).

Journal of Membrane Science 533 (2017) 201–209

Available online 31 March 20170376-7388/ © 2017 Elsevier B.V. All rights reserved.

MARK

ally coating the polystyrene-block-polylactide (PS-b-PLA) solution onpolyethersulfone microfiltration (MF) membranes with pores pre-viously filled with water, followed by selective etching the sacrificialPLA blocks [30]. Recently, we manually coated dilute polystyrene-block-poly (2-vinylpyrdine) (PS-b-P2VP) solutions on water-filled poly-vinylidenefluoride (PVDF) MF membranes, and immersed the coatedmembranes into hot ethanol for hours to generate nanoporosities in theBCP layers by the selective swelling mechanism, thus producing BCP/PVDF composite membranes delivering an ultrafiltration (UF) function[21,28]. However, the direct coating method is predominantly using UFor MF membranes as the macroporous supports which are much lesspermeable and more expensive than nonwoven extensively used assupports for UF and MF membranes. Moreover, the composite mem-branes are only available with small areas because coating of BCPsolutions on porous substrates is typically manually performed withpoor controllability. These issues significantly hamper the reproducibleproduction of BCP membranes in large scale and their usages in real-world applications.

Polyester nonwoven fabrics can be cheaply sourced and they aremechanical strong and ductile, and also highly porous and waterpermeable. Therefore, they might serve as a good macroporoussubstrate to support nanoporous BCP layer. However, their pore sizesare much larger than that of UF/MF membranes typically used assupport for BCP membranes, and their surface structure and chemistryare also significantly different. We notice that polyester nonwoven hasbeen directly used to support BCP membranes produced through theNIPS process in which highly concentrated BCP solutions are involved[31,32]. However, polyester nonwoven has not been used in other twomethods (selective etching and selective swelling) because they startfrom much thinner BCP solutions which easily leak out of thenonwoven. Therefore, it is necessary to have a focused study to revealthe feasibility and possible merits of production of swelling-resultantBCP membranes using nonwoven as the support. To this end, we firstmachine-cast PS-b-P2VP solutions on the nonwoven fabrics followed bythermal treatment, and cavitate the BCP layers by selective swelling,thus producing composite membranes with nanoporous BCP as theselective layers and macroporous nonwoven as the supporting sub-strates. Thanks to the controllable machine-casting of BCP solutions onpolyester nonwovens, thus-produced BCP composite membranes ex-hibit excellent permselectivity tunable by altering the swelling tem-peratures and/or durations. Moreover, we demonstrate that the mem-branes can be used to fractionate nanoparticles by reducing their sizedistributions.

2. Experimental section

2.1. Materials

The block copolymer of PS-b-P2VP [Mn (PS)=53,000 g mol−1, Mn

(P2VP)=21,000 g mol−1, polydispersity index (PDI)=1.17] was ob-tained from Tubang Polymer Materials Co., Ltd. The polyester non-woven (E055094-74) was purchased from Suzhou Holykem AutomaticTechnology Co., Ltd and used as received. Organic solvents includingchloroform and ethanol with analytical grade were sourced from localsuppliers. The protein, bovine serum albumin (BSA) with the molecularweight of 67 kDa and a negatively charged dye, Acid Orange 7, werepurchased from Aladdin Industrial Corporation. Monodispersed goldcolloidal nanoparticles with the diameter of 15 nm were obtained fromBritish Biocell International, and polydispersed gold nanoparticles withthe size in the range of ~3–20 nm were purchased from ShanghaiHuzheng Nanotechnology Co., Ltd. All the chemical reagents were usedwithout further purification.

2.2. Membrane fabrication

The PS-b-P2VP BCP was first dissolved in chloroform at a concen-

tration of 10 wt%. A clean and dust-free glass plate was then placedonto the horizontal surface of the casting machine (AFA-II, ShanghaiXiandai Environmental Engineering Technique Co., Ltd). A piece ofnonwoven with the size of 15 cm×20 cm, used as the supporting layer,was immersed in deionized water for about 5 min, allowing water topercolate through the pores in it. The nonwoven was then withdrawnfrom the water bath, gently shaken to remove excessive water andplaced on the top of the glass plate. Afterwards, the BCP solution wasmachine-cast onto the nonwoven by using a casting knife with~200 µm gate height to evenly spread the solution over the substrate.The as-cast nonwoven was then kept in the fume hood at roomtemperature for about 6 h to remove both the residual organic solventand the water filled in the pores, followed by heating in vacuum at100 °C for another 1 h to dry the membrane. Evaporation of the solventled to a thin and dense BCP layer on the top of the nonwoven,producing a bi-layered composite structure. To generate pores in theBCP layer, the selective swelling process was applied [10,25]. Briefly,the BCP-coated nonwoven was immersed in warm ethanol for desiredperiods of time, followed by air drying at room temperature.

2.3. Characterizations

The morphological features of composite membranes were exam-ined with a Hitachi S-4800 field emission scanning electron microscope(SEM) operated at 5 kV. In the preparation of the samples for cross-sectional examination, the membranes were fractured in liquid nitro-gen. Prior to SEM observations, the samples were sputter-coated with athin layer of platinum to reinforce their conductivity. A contact anglegoniometer (DropMeter A100, Maist) was used to measure the dynamicwater contact angles (WCAs) of the BCP layers coated on the nonwovenbefore and after swelling, and the test time was fixed at 3 min. For eachsample, at least five randomly chosen locations of the sample surfacewere measured and the mean value was reported. The size-distributioncurves of gold nanoparticles in the feed and filtrate were measured bydynamic laser scattering method (Nano-ZS90, Malvern).

2.4. Filtration tests

Membrane coupons with the diameter of 2.5 cm were cut from largemembrane sheets and were used for the filtration tests. All the filtrationtests were carried out on a stirred filtration cell (Amicon 8010,Millpore) with a 10 mL working volume and an effective membranearea of 4.1 cm2. For the tests of pure water permeability (PWP), eachmembrane was initially pressurized in deionized water for 10 min at0.5 bar to ensure a stabilized permeability, and then the PWP wasmeasured at 0.2 bar. The hydraulic permeability of the membrane wasdetermined by the ratio of the volumetric filtrate flux (L m−2 h−1) tothe trans-membrane pressure (bar). To measure separation properties,BSA was first dissolved in the phosphate buffered solution at aconcentration of 0.5 g L−1, which was then used as the feed solution.Concentrations of the feed and filtrate solutions were measured with aUV–vis absorption spectrometer (NanoDrop 2000c, Thermo Scientific)at ~280 nm. Aqueous solutions of colloidal gold nanoparticles withmonodispersed size of 15 nm were also used as the feed solutions to testthe retention properties of membranes prepared under different swel-ling conditions. To eliminate any adsorption effects of gold nanoparti-cles on the membrane surface, we followed our previous work [33] andconditioned the composite membranes in anionic Acid Orange 7 for30 min to form a thin adsorption layer on the membrane, making themembrane negatively charged. Therefore, the adsorption of the nega-tively charged gold nanoparticles on the membrane can be neglected.Gold nanoparticles with polydispersed particle sizes in the range of~3–20 nm were used to test the size fractionation capability of themembrane prepared by swelling at 60 °C for 3 h. The gold concentra-tions in feed, filtrate and retentate were also determined with theUV–vis absorption spectrometer at ~520 nm.

X. Shi et al. Journal of Membrane Science 533 (2017) 201–209

202

3. Results and discussion

3.1. Preparation of nanoporous BCP membranes on nonwoven supports

Fig. 1 shows the schematic diagram for the fabrication process of thePS-b-P2VP composite membranes supported on the nonwoven. Thewater prefilled nonwoven was first obtained and placed onto the glassplate (Fig. 1a). The 10 wt% PS-b-P2VP solution was then machine-castonto the nonwoven to obtain a liquid BCP layer (Fig. 1b). After solventevaporation and air drying, a thin and dense BCP layer was formed onthe top of the nonwoven with an area of ~150 cm2 (Fig. 1c),demonstrating the efficiency and scalability of this method. The SEMresults present significant differences between the original nonwovenand the BCP-coated nonwoven (Fig. S1). The feature size of the voids inthe nonwoven is ranging from 2 µm to 30 µm, and a dense BCP layercan be observed right on the support after drying. The BCP-coatednonwoven was then immersed in warm ethanol for various durations tocavitate the BCP layer by the selective swelling-induced pore genera-tion process. However, early attempts to generate pores were plaguedby macroporous cracks in the BCP layer during the swelling processbecause of the residual stress in the coated BCP layers [30,34]. Toaddress this problem, pretreatment of heating in vacuum at 100 °C for1 h followed by natural cooling to room temperature was introducedbefore swelling. During heating, the mobility of the polymer chains aresignificantly enhanced as the temperature is near the glass transitiontemperature (Tg) of both PS [35] and P2VP [36] blocks, so as to relaxthe residual stress [37]. The subsequent cooling minimizes the stresscaused by the difference in the coefficients of thermal expansionbetween the BCP layer and the nonwoven support. Moreover, theheating treatment is also helpful to enhance the adhesion between theBCP layer and the nonwoven support. After such a heating treatment,bicontinuous nanoporosities can be successfully generated in the BCPlayer without any cracking after ethanol swelling (Fig. 1d). Thecavitation of the BCP selective layer follows the mechanism of selectiveswelling (magnified illustration in Fig. 1). Upon immersion of PS-b-P2VP in ethanol, ethanol diffuses into the BCP layer and is preferen-tially enriched in the P2VP microdomains as ethanol is a good solventto P2VP but a nonsolvent to PS. The P2VP microdomains are accord-ingly swollen and expanded to merge with their neighbors, resulting ina continuous phase distributed in the PS matrix. Upon drying, theswollen P2VP chains collapse with the evaporation of ethanol, while the

spaces initially occupied by the expanding P2VP chains are immobi-lized because of the PS matrix which is in the glassy state. Conse-quently, bicontinuous porosity is generated in the BCP layer. We notethat the BCP-coated nonwoven maintained good structural integrityand no cracks or peeling-off can be observed after the swellingtreatment. Meanwhile, the BCP-coated side of the nonwoven exhibiteda uniform smooth and shining surface while the opposite side retainedthe initially dull appearance after the coating, heating, and swellingtreatment.

3.2. Morphology of nonwoven-supported nanoporous BCP membranes

We investigated the morphologies of the membranes preparedunder different swelling conditions. Fig. 2a–e show the surfacemorphologies of the membranes prepared by ethanol swelling at60 °C for various durations from 15 min to 12 h. As can be seen fromFig. 2a, a swelling duration as short as 15 min is sufficient to introducepores into the coated BCP layers. These pores are mainly present in twodifferent geometries: circular pores and narrow channels. The coex-istence of the two geometries of pores are reminisces of the perpendi-cularly and in-plane oriented P2VP cylinders embedded in the PSmatrix before swelling, because the BCP coating layers were notsubjected to any alignment and the P2VP microdomains were randomlydistributed in the PS matrix [38]. These two types of pores are presentin all membranes prepared under different swelling conditions.Although the pore size is difficult to be accurately determined, thereis a clear evident trend that the pores are enlarged with prolongedswelling durations at the same swelling temperature. As can be seenfrom Table S1, the average pore size of the membranes and the standarddeviation (σ) values were obtained by evaluation of minimum 100nanopores randomly selected in SEM images using the software NanoMeasurer (for channel-like pores, the pore width is considered to be thepore diameter) [24]. The average pore size in Fig. 2a is 13.9 nm(σ=±3.2 nm) whereas the average pore size of the membraneprepared by swelling at 60 °C for 12 h is 21.2 nm (σ=±4.3 nm)(Fig. 2e). In addition, the surface morphology of the BCP layer becomesincreasingly rough after longer swelling durations due to a highercondition of swelling, which leads to the drastic volume expansion andoverflow of the P2VP chains. A moderate increase of the swellingtemperature to 65 °C noticeably enlarges the pore sizes, and at thisswelling temperature the increase in pore diameters with swelling

Fig. 1. Schematic diagram for the fabrication process of the nanoporous BCP membranes machine-cast on the nonwoven support. The drying step contains solvent evaporation at roomtemperature and heating treatment at 100 °C.

X. Shi et al. Journal of Membrane Science 533 (2017) 201–209

203

Fig. 2. Surface SEM images of nanoporous BCP layers machine-cast on nonwoven supports after ethanol swelling under different conditions. (a-e) Swelling at 60 °C for 15 min, 1 h, 3 h,6 h, and 12 h, respectively. (f-j) Swelling at 65 °C for 15 min, 1 h, 3 h, 6 h, and 12 h, respectively.

Fig. 3. SEM images of the membrane prepared by ethanol swelling at 60 °C for 6 h. (a) Large-field view of the membrane surface, (b and c) Cross-sectional view of the membrane. (c) isthe corresponding magnified images of the boxed areas in (b).

X. Shi et al. Journal of Membrane Science 533 (2017) 201–209

204

durations becomes more pronounced (Fig. 2f–j). For instance, theaverage pore size is enlarged from 20.4 nm (σ=±4.3 nm) to33.4 nm (σ=±8.2 nm) with the swelling durations increased from15 min to 12 h. By comparing the membranes prepared with the sameswelling durations but at different temperatures, it is clear that swellingat 65 °C is always producing larger pores than swelling at 60 °C. Thiscan be easily understood as higher temperatures enhance both thesolvation of the P2VP chains and the mobility of the PS matrix inethanol, leading to higher osmotic pressure accumulated in the P2VPmicrodomains confined in the PS matrix which has stronger deform-ability. Consequently, within an identical swelling duration larger poresare generated in the membrane prepared at higher swelling tempera-tures.

Fig. 3a gives a large-field view of the surface of the membraneprepared by swelling at 60 °C for 6 h, and it is clear that the porousmorphology with interconnected porosity exists homogeneouslythroughout the whole area of the BCP surface. Moreover, as can beseen from Fig. 3b, the membrane possesses a double-layered compositestructure with a thin, nanoporous top layer supported on a thick,macroporous nonwoven bottom layer. Clearly, the top layer is aswelling-induced PS-b-P2VP layer with a thickness of ~10–15 µmuniformly adhered on the nonwoven. Magnified SEM examinations onthe cross section of the membrane show interconnected porosityrunning from the top surface to the bottom surface, and we do notnotice a gradient in porosity along the cross section.

3.3. Surface wettability of nonwoven-supported nanoporous BCPmembranes

We prepared porous membranes with the BCP selective layerssubjected to ethanol swelling under various conditions. With the

change of microstructure and surface chemical constitution afterswelling, the advantage of improved wettability came along. Hence,the WCAs of composite membranes before and after the treatment weremeasured. Fig. 4a presents the typical WCA curves of compositemembranes with and without swelling. For the pristine sample (withoutswelling), an initial WCA of ~87° was measured. This value is close tothat of PS homopolymers [39], implying that the surfaces of BCP layersare preferentially enriched with PS blocks after casting. Meanwhile, theWCA barely changes during a measurement period of 1 min (insets inFig. 4a for the pristine membrane) implying a complete coating layer onthe macroporous nonwoven support. However, as for the membranessubjected to swelling at 60 °C/65 °C for 3 h, the initial WCA decreasesto ~60° and ~54°, respectively. The results indicate that the surfacechemical composition changes due to the enrichment of the hydrophilicP2VP blocks on the surface after swelling, which was confirmed in ourprevious work by X-ray photoelectron spectroscopy [21]. Furthermore,the water droplet on membrane surface quickly penetrates into themembrane after 1 min time interval (insets in Fig. 4a for the swollenmembranes), demonstrating the presence of accessible pores in the BCPlayer after swelling.

For further comparison, the WCAs for membranes prepared undervarious swelling conditions were tested. To have a direct comparisonfor the change of WCA of different membranes, the requisite time forthe WCA decreased to 40° of each membrane was recorded. As can beseen from Fig. 4b, the requisite time decreases steadily with theswelling duration for membranes prepared at both 60 °C and 65 °C, asthe BCP layers become more porous and more P2VP chains migrate tothe surface as the swelling goes on. Meanwhile, the requisite time forthe membranes prepared by ethanol swelling at 65 °C is always muchshorter than that of the membranes prepared by swelling at 60 °C forthe same duration. This can be ascribed to the pronounced poregeneration and the drastic increase of surface roughness when theswelling temperature is increased to 65 °C. The hydrophilicity of theBCP layer is enhanced with higher surface roughness according to theWenzel state [40]. As a shorter requisite time implies that themembrane possesses a better wettability, Fig. 4b also reveals thatelevation in the swelling temperature is more effective to increase thewettability of the membrane than the extension of the swellingduration.

3.4. Permselectivity of nonwoven-supported nanoporous BCP membranes

As discussed in the last section, the hydrophilic surface, the porousstructure, and the increasing roughness contribute to improve thesurface hydrophilicity of the membranes after swelling, which is helpfulto improve the water permeation. Herein, the permeability of the

Fig. 4. (a) Dynamic water contact angles of the BCP membranes prepared under differentswelling conditions. (b) The requisite time for water contact angles of the BCP membranesdropped to 40°.

Fig. 5. The pure water permeability of membranes prepared by swelling in ethanol at60 °C and 65 °C for various durations.

X. Shi et al. Journal of Membrane Science 533 (2017) 201–209

205

membranes prepared under various swelling conditions is given inFig. 5. Note that prior to swelling the BCP membrane was impermeableto water under 0.5 bar, indicating the nonporous and dense nature ofthe as-coated BCP layers. In contrast, the membrane shows a slight PWPof 33 L m−2 h−1 bar−1 after swelling at 60 °C for only 15 min, whichimplies formation of accessible pores in the BCP layer allowing thepermeability of water throughout the membrane. After swelling for 1 h,the PWP is increased to 88 L m−2 h−1 bar−1. As expected, withextending the swelling duration from 3 through 6 to 12 h, themembrane exhibits enhanced permeability from 152 through 202 to309 L m−2 h−1 bar−1, respectively. Impressively, remarkable increasein the PWP is observed when the swelling temperature is increased to65 °C. For instance, The PWP is greatly increased to106 L m−2 h−1 bar−1 after ethanol swelling at 65 °C for 15 min. ThePWP is continuously increased to ~850 L m−2 h−1 bar−1 when theswelling at 65 °C was prolonged to 12 h. Clearly, the PWP curvesdisplay an increasing tendency with the extension of swelling duration,which coincides with WCA results. However, the increase in PWP is notlinear with the swelling duration. As can be seen in Fig. 5, PWP isincreased faster in the first 1 h and 3 h for the membrane prepared byswelling at 60 °C and 65 °C, respectively, and continues to increase butat a lower rate. The nonlinear increase in PWP with swelling durationsis believed to be due to the opposite effect of water diffusibility andmembrane thickness. In the initial stage of swelling, water diffusibilitythrough the membrane is playing the dominated role in determiningPWPs because the pore size and surface hydrophilicity are dramaticallyincreased. With prolonged swelling, the surface hydrophilicity remainsalmost unchanged and the enlargement in pore size also turns to be lesspronounced. Thus, a tradeoff effect would appear as the increasedmembrane thickness would take the leading role because of greatermass-transfer resistance, resulting in reduced increase in PWP. Further-more, membranes prepared by swelling at 65 °C present larger poresizes compared to membranes prepared by swelling at 60 °C and thePWP is proportional to the pore diameter. Therefore, the compositemembranes subjected to swelling at a higher temperature require alonger swelling duration to achieve the shift point.

We then investigated the separation performance of the machine-cast BCP membranes fabricated under different swelling conditions.BSA and monodispersed gold nanoparticles with the size of 15 nm werechosen as two model materials to probe the size-sieving separationcapability of the membranes. As can be seen from Fig. 6, all thesemembranes exhibit modest retention rates ranging from 50% to 10% toBSA, and the retention rate declines with the increase of both swellingdurations and swelling temperatures as a result of enlarged pore sizes,which is opposite to the change of PWP as discussed above. In contrast,when 15-nm gold nanoparticles were used, the membranes prepared byswelling at 60 °C for various durations exhibit a similar retention of~95%. Such a high retention to 15-nm nanoparticles reveals that themembranes are free of defects and the effective pore size of themembranes is lower than 15 nm. As shown in Fig. 2 the pore sizescharacterized by SEM are typically larger than 15 nm, however, theeffective pore sizes for the membranes used in aqueous systems areexpected to be reduced because of the swelling of P2VP chains enrichedon the pore wall in water. The BSA protein is reported to be in aellipsoidal shape with the size of 14 nm×3.8 nm×3.8 nm [41], whilethe shape of the gold nanoparticles are in the spherical shape. Duringthe rejection tests, the isotropic gold nanoparticles are more effective tobe intercepted by the bicontinuous pores in the BCP layers, while theBSA molecules have more chance to percolate through the pores bychanging their orientation toward the pores [42], leading to modestBSA retentions. For the membranes prepared by swelling at 65 °C theretention to 15-nm gold nanoparticles is decreased from ~93% to~54% when the swelling duration is increased from 15 min to 12 h as aresult of increased pore sizes. It is worth noting that the swellingduration increased from 15 min to 12 h leads to decreased retention to15-nm gold nanoparticles (< 50%), however, this decrease in retention

is greatly paid back by remarkable increase in the PWP for nearly 8times (Fig. 5). Moreover, Fig. 5 and 6 suggest that adjustable separatingproperties in wide ranges can be achieved simply by altering theswelling conditions including both swelling durations and tempera-tures. The permeabilities of the membranes prepared by swelling at60 °C for different durations to solutions of 15-nm gold nanoparticleswere also recorded and compared with their pure water permeabilities.As can be seen from Fig. 7, the permeability of gold solutions isapproximately 12–40% lower than PWP for all the membranes.Considering that gold nanoparticles hamper the transport of wateracross the membrane pores such reduction in permeabilities is accep-table.

Fig. 6. The retention rates to BSA and 15-nm gold nanoparticles of the membranesprepared under various swelling conditions.

Fig. 7. The permeability of pure water and the solution containing 15-nm goldnanoparticles across the membrane prepared by swelling at 60 °C for various durations.

X. Shi et al. Journal of Membrane Science 533 (2017) 201–209

206

3.5. Mechanical stability of nonwoven-supported nanoporous BCPmembranes

As the nanoporous BCP layers are only about ten micrometers inthickness and the nonwoven supports possess pores with sizes up to tensof micrometers we should demonstrate that thus produced compositemembranes are sufficiently strong for real applications. We first testedthe pressure-dependent PWPs of the membrane prepared by ethanolswelling at 60 °C for 6 h. As shown in Fig. 8, the PWP is linearlyincreased with the trans-membrane pressure, and neither significantdrop nor abrupt increase of the PWP is observed when the pressure risesup to 2.5 bar. It is obvious that the good resistance to pressures shouldbe attributed to the robust nonwoven substrate as well as theinterconnected porosities in the BCP layer which help to dissipatestresses. The influence of test duration on the PWP behavior was alsoinvestigated with results given in Fig. 9. As for the membrane preparedby swelling at 60 °C for 6 h, the PWP shows no noticeable drop within1 h of filtration test. However, for the membrane prepared by swellingat 65 °C for 6 h the PWP is decreased by about 20% after filtration forthe first 20 min. The permeability decline is mainly caused by thecompaction of the membrane. The membrane prepared at higherswelling temperature possesses larger pore sizes and higher porosities(insets in Fig. 9), and therefore, tolerate pressures to a less degree.However, the PWP is soon stabilized with continuing filtration andmaintains at high level of ~710 L m−2 h−1 bar−1. Therefore, the

pressure-dependent and time-dependent PWP results indicate that thusproduced composite membranes possess a reasonably good mechanicalstability and durability.

3.6. Fractionation of nanoparticles by nonwoven-supported nanoporousBCP membranes

Two general strategies have been employed to produce size-uniformnanoparticles during the practical synthesis process. One method is thedirect particle size management during the synthesis by adjustingfabrication parameters [43], the other is the post-fractionation withthe membrane from the polydispersed nanoparticle samples [44]. Bytaking advantage of the relatively narrow pore size distribution in theBCP layer, the composite membranes may be used for the sizefractionation of nanoparticles. As demonstrated in the filtration ofprotein solutions and gold nanoparticles, the composite membranesshowed a tunable size-sieving performance. Furthermore, we exploredtheir applications in the fractionation of nanoparticles. A colloidalsolution of polydispersed gold nanoparticles with sizes ranging from ~3to 20 nm was used to challenge the membrane prepared by ethanolswelling at 60 °C for 3 h. Fig. 10a displays the UV–vis spectra as well asthe color appearance (inset) of the feed, filtrate, and retentate involvedin the fractionation test. A moderate absorption peak around 530 nmcan be observed in the UV–vis spectrum of the feed solution and thepeak is intensified in the spectrum of the retentate, while an ignorablepeak exists in the spectrum of the filtrate. These results indicate that themembrane exhibits a certain retention to the particles. The difference incolor appearance of the three solutions also directly reveals a noticeable

Fig. 8. The pressure-dependent water flux of the membrane prepared by ethanol swellingat 60 °C for 6 h.

Fig. 9. The time-dependent water permeability of composite membranes prepared byethanol swelling for 6 h at 60 °C and 65 °C, respectively. Insets are cross-sectional SEMimage of the corresponding membranes.

Fig. 10. Nanoparticles fractionation with the membrane prepared by swelling at 60 °C for3 h. (a) The UV–vis spectra of the feed, filtrate, and retentate of polydispersed goldnanoparticles. Inset shows the picture of the three solutions displaying various colorappearance. (b) The size distributions of the feed and filtrate solutions collected from thesize-fractionation of ~3–20 nm gold nanoparticles. (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article).

X. Shi et al. Journal of Membrane Science 533 (2017) 201–209

207

retention of the polydispersed gold nanoparticles. Note that both theabsence of the peak and colorless appearance in the filtrate do notnecessarily imply that the retention solution is free of any goldparticles, as the gold particles with sizes lower than 10 nm do not givea characteristic adsorption peaks because of their weak plasmoniceffect. To further investigate the existence of the particles in the filtrate,the filtrate solution was tested by dynamic laser scattering. As shown inFig. 10b, the feed solution contains nanoparticles with a wide sizedistribution in the range of ~3–20 nm, which is confirmed by the curveof the feed. A small amount of particles with the size of ~30 nm canalso be observed due to the aggregation of some small gold nanopar-ticles. However, a much narrow size distribution ranging from ~3 to10 nm was measured in the filtration, whereas the large particles arerepelled. After filtration, the mean particle size is reduced from ~7 nmin feed to ~5 nm in the filtration. The results imply the membrane isable to narrow down the size distribution of nanoparticles, and a moreprecise fractionation is expected by using a series of membranes withtunable effective pore sizes.

4. Conclusions

In summary, we explore the fabrication of nanoporous BCP compo-site membranes via mechanically casting amphiphilic BCP solutionsover robust nonwoven supports. Solvent evaporation followed byswelling-induced pore generation results in nanoporous size-sievinglayers with the thickness of about ten micrometers tightly adhered tononwoven supports. Changes in temperature and duration during theselective swelling allow flexible regulation of porous structures as wellas surface properties, leading to tunable separation performances. Thecomposite membranes are also demonstrated to be mechanically stableand durable in pressurized filtrations. Moreover, the membranes canefficiently fractionate polydispersed nanoparticles to narrow downtheir size distribution. This convenient and controllable strategy offersthe possibility for the production of composite nanoporous BCPmembranes with good performance, low cost, and excellent upscal-ability.

Acknowledgements

Financial support from the National Basic Research Program ofChina (2015CB655301), the Jiangsu Natural Science Foundation(BK20150063), and the Project of Priority Academic ProgramDevelopment of Jiangsu Higher Education Institutions (PAPD) is grate-fully acknowledged.

Appendix A. Supporting information

Supplementary data associated with this article can be found in theonline version at doi:10.1016/j.memsci.2017.03.046.

References

[1] M. Elimelech, W.A. Phillip, The future of seawater desalination: energy, technology,and the environment, Science 333 (2011) 712–717.

[2] C. Charcosset, Membrane processes in biotechnology: an overview, Biotechnol.Adv. 24 (2006) 482–492.

[3] E. Gultepe, D. Nagesha, S. Sridhar, M. Amiji, Nanoporous inorganic membranes orcoatings for sustained drug delivery in implantable devices, Adv. Drug Deliv. Rev.62 (2010) 305–315.

[4] H.J. Kim, K. Choi, Y. Baek, D.G. Kim, J. Shim, J. Yoon, J.C. Lee, High-performancereverse osmosis CNT/polyamide nanocomposite membrane by controlled inter-facial interactions, ACS Appl. Mater. Interfaces 6 (2014) 2819–2829.

[5] W.Z. Li, X. Wang, Z.W. Chen, M. Waje, Y.S. Yan, Carbon nanotube film by filtrationas cathode catalyst support for proton-exchange membrane fuel cell, Langmuir 21(2005) 9386–9389.

[6] G.P. Liu, W.Q. Jin, N.P. Xu, Graphene-based membranes, Chem. Soc. Rev. 44 (2015)5016–5030.

[7] X. Zhao, Y. Su, Y. Liu, Y. Lip, Z. Jiang, Free-standing graphene oxide-palygorskitenanohybrid membrane for oil/water separation, ACS Appl. Mater. Interfaces 8(2016) 8247–8256.

[8] Y.X. Hu, X.L. Dong, J.P. Nan, W.Q. Jin, X.M. Ren, N.P. Xu, Y.M. Lee, Metal-organicframework membranes fabricated via reactive seeding, Chem. Commun. 47 (2011)737–739.

[9] H.B.T. Jeazet, C. Staudt, C. Janiak, Metal-organic frameworks in mixed-matrixmembranes for gas separation, Dalton Trans. 41 (2012) 14003–14027.

[10] Y. Wang, Nondestructive creation of ordered nanopores by selective swelling ofblock copolymers: toward homoporous membranes, Acc. Chem. Res. 49 (2016)1401–1408.

[11] L. Wang, S.L. Ji, N.X. Wang, R. Zhang, G.J. Zhang, J.R. Li, One-step self-assemblyfabrication of amphiphilic hyperbranched polymer composite membrane fromaqueous emulsion for dye desalination, J. Membr. Sci. 452 (2014) 143–151.

[12] E.A. Jackson, M.A. Hillmyer, Nanoporous membranes derived from block copoly-mers: from drug delivery to water filtration, ACS Nano 4 (2010) 3548–3553.

[13] Y. Wang, F. Li, An emerging pore-making strategy: confined swelling-induced poregeneration in block copolymer materials, Adv. Mater. 23 (2011) 2134–2148.

[14] K.-V. Peinemann, V. Abetz, P.F.W. Simon, Asymmetric superstructure formed in ablock copolymer via phase separation, Nat. Mater. 6 (2007) 992–996.

[15] A. Bertrand, M.A. Hillmyer, Nanoporous poly(lactide) by olefin metathesis degra-dation, J. Am. Chem. Soc. 135 (2013) 10918–10921.

[16] E.A. Jackson, Y. Lee, M.A. Hillmyer, ABAC tetrablock terpolymers for toughnanoporous filtration membranes, Macromolecules 46 (2013) 1484–1491.

[17] H. Sai, K.W. Tan, K. Hur, E. Asenath-Smith, R. Hovden, Y. Jiang, M. Riccio,D.A. Muller, V. Elser, L.A. Estroff, S.M. Gruner, U. Wiesner, Hierarchical porouspolymer scaffolds from block copolymers, Science 341 (2013) 530–534.

[18] Z. Yi, P.-B. Zhang, C.-J. Liu, L.-P. Zhu, Symmetrical permeable membranesconsisting of overlapped block copolymer cylindrical micelles for nanoparticle sizefractionation, Macromolecules 49 (2016) 3343–3351.

[19] M. Radjabian, V. Abetz, Tailored pore sizes in integral asymmetric membranesformed by blends of block copolymers, Adv. Mater. 27 (2015) 352–355.

[20] R.M. Dorin, H. Sai, U. Wiesner, Hierarchically porous materials from blockcopolymers, Chem. Mater. 26 (2014) 339–347.

[21] Z. Wang, X. Yao, Y. Wang, Swelling-induced mesoporous block copolymermembranes with intrinsically active surfaces for size-selective separation, J. Mater.Chem. 22 (2012) 20542–20548.

[22] M. Wei, W. Sun, X. Shi, Z. Wang, Y. Wang, Homoporous membranes with tailoredpores by soaking block copolymer/homopolymer blends in selective solvents:dissolution versus swelling, Macromolecules 49 (2016) 215–223.

[23] H. Ahn, S. Park, S.W. Kim, P.J. Yoo, D.Y. Ryu, T.P. Russell, Nanoporous blockcopolymer membranes for ultrafiltration: a simple approach to size tunability, ACSNano 8 (2014) 11745–11752.

[24] X. Yao, L. Guo, X. Chen, J. Huang, M. Steinhart, Y. Wang, Filtration-based synthesisof micelle-derived composite membranes for high-flux ultrafiltration, ACS Appl.Mater. Interfaces 7 (2015) 6974–6981.

[25] W. Sun, Z. Wang, X. Yao, L. Guo, X. Chen, Y. Wang, Surface-active isoporousmembranes nondestructively derived from perpendicularly aligned block copoly-mers for size-selective separation, J. Membr. Sci. 466 (2014) 229–237.

[26] S.Y. Yang, I. Ryu, H.Y. Kim, J.K. Kim, S.K. Jang, T.P. Russell, Nanoporousmembranes with ultrahigh selectivity and flux for the filtration of viruses, Adv.Mater. 18 (2006) 709–712.

[27] L. Guo, Y. Wang, Nanoslitting of phase-separated block copolymers by solventswelling for membranes with ultrahigh flux and sharp selectivity, Chem. Commun.50 (2014) 12022–12025.

[28] Z. Wang, Y. Wang, Highly permeable and robust responsive nanoporous membranesby selective swelling of triblock terpolymers with a rubbery block, Macromolecules49 (2016) 182–191.

[29] S.E. Querelle, E.A. Jackson, E.L. Cussler, M.A. Hillmyer, Ultrafiltration membraneswith a thin poly(styrene)-b-poly(isoprene) selective layer, ACS Appl. Mater.Interfaces 5 (2013) 5044–5050.

[30] W.A. Phillip, B. O’Neill, M. Rodwogin, M.A. Hillmyer, E.L. Cussler, Self-assembledblock copolymer thin films as water filtration membranes, ACS Appl. Mater.Interfaces 2 (2010) 847–853.

[31] S.P. Nunes, R. Sougrat, B. Hooghan, D.H. Anjum, A.R. Behzad, L. Zhao, N. Pradeep,I. Pinnau, U. Vainio, K.-V. Peinemann, Ultraporous films with uniform nanochan-nels by block copolymer micelles assembly, Macromolecules 43 (2010) 8079–8085.

[32] S.P. Nunes, A.R. Behzad, K.-V. Peinemann, Self-assembled block copolymermembranes: from basic research to large-scale manufacturing, J. Mater. Res. 28(2013) 2661–2665.

[33] L. Guo, L. Wang, Y. Wang, Stretched isoporous composite membranes with ellipticnanopores for external-energy-free ultrafiltration, Chem. Commun. 52 (2016)6899–6902.

[34] X. Zhang, J.F. Douglas, R.L. Jones, Influence of film casting method on blockcopolymer ordering in thin films, Soft Matter 8 (2012) 4980–4987.

[35] M.B. Yaffe, E.J. Kramer, Plasticization effects on environmental craze microstruc-ture, J. Mater. Sci. 16 (1981) 2130–2136.

[36] X.Q. Jiang, C.Z. Yang, K. Tanaka, A. Takahara, T. Kajiyama, Effect of chain endgroup on surface glass transition temperature of thin polymer film, Phys. Lett. A 281(2001) 363–367.

[37] X. Zhang, S.H.D.P. Lacerda, K.G. Yager, B.C. Berry, J.F. Douglas, R.L. Jones,A. Karim, Target patterns induced by fixed nanoparticles in block copolymer films,ACS Nano 3 (2009) 2115–2120.

[38] Y. Wang, L. Tong, M. Steinhart, Swelling-induced morphology reconstruction inblock copolymer nanorods: kinetics and impact of surface tension during solventevaporation, ACS Nano 5 (2011) 1928–1938.

[39] P. Mansky, Y. Liu, E. Huang, T.P. Russell, C.J. Hawker, Controlling polymer-surfaceinteractions with random copolymer brushes, Science 275 (1997) 1458–1460.

[40] C. Ran, G. Ding, W. Liu, Y. Deng, W. Hou, Wetting on nanoporous alumina surface:

X. Shi et al. Journal of Membrane Science 533 (2017) 201–209

208

transition between Wenzel and Cassie states controlled by surface structure,Langmuir 24 (2008) 9952–9955.

[41] C.C. Striemer, T.R. Gaborski, J.L. McGrath, P.M. Fauchet, Charge- and size-basedseparation of macromolecules using ultrathin silicon membranes, Nature 445(2007) 749–753.

[42] X. Qiu, H. Yu, M. Karunakaran, N. Pradeep, S.P. Nunes, K.-V. Peinemann, Selective

separation of similarly sized proteins with tunable nanoporous block copolymermembranes, ACS Nano 7 (2013) 768–776.

[43] X. Wang, Y. Li, Monodisperse nanocrystals: general synthesis, assembly, and theirapplications, Chem. Commun. (2007) 2901–2910.

[44] S.F. Sweeney, G.H. Woehrle, J.E. Hutchison, Rapid purification and size separationof gold nanoparticles via diafiltration, J. Am. Chem. Soc. 128 (2006) 3190–3197.

X. Shi et al. Journal of Membrane Science 533 (2017) 201–209

209